The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell and Python scripts for operational tasks including checking external skills and managing the skill lifecycle. The file 'hooks/external-skills-check.sh' is a wrapper that identifies the root directory and executes 'scripts/check-external-skills.py' with passed arguments. Documentation in 'SKILL.md' and 'references/routing-guide.md' identifies scripts like 'install-external-skills.py', 'skillctl.py', and 'validate_skills.py' as primary tools for skill management.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of additional capabilities from external sources as part of its core functionality. The 'README.md' file contains an installation example using 'npx -y skills add', which downloads packages from the npm registry. The 'SKILL.md' file defines a list of 'External Skill Chains' and provides corresponding installation commands that point to the vendor's scripts.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection (Category 8) due to its role in auditing and processing the instructions of other skills. Ingestion points: 'references/routing-guide.md' describes an audit checklist that involves reading and validating 'SKILL.md' and '.skillmeta.yml' files from other skills. Boundary markers: The current documentation does not define specific delimiters for separating audited content from the agent's primary instructions. Capability inventory: The skill environment has access to file-system operations and script execution (e.g., python3, bash, npx) used for skill maintenance. Sanitization: There is no evidence in the provided files of explicit sanitization or validation of the content being audited.

skills-management